The National AI Research Resource (NAIRR) is a proposed and pilot-stage United States initiative designed to expand access to computing infrastructure, datasets, models, and educational tools for artificial intelligence research. Its primary goal is to democratize AI research by providing resources that are currently concentrated in a few large technology companies and elite universities, thereby enabling a broader community of researchers, students, and small businesses to participate in AI development. The initiative is a collaborative effort involving federal agencies, private sector partners, and academic institutions, and it operates under the guidance of the White House Office of Science and Technology Policy (OSTP) and the National Science Foundation (NSF).
The concept of a shared national research infrastructure for AI emerged from a 2020 report by the National Security Commission on Artificial Intelligence, which recommended creating a federated cloud-based resource. In January 2023, the OSTP and NSF released a roadmap for implementing the NAIRR, and a pilot program was launched in January 2024. The pilot aims to test the feasibility of a full-scale resource, with an initial investment of approximately $200 million from federal agencies and contributions from private sector partners.
Pilot Program and Partnerships
The NAIRR pilot, announced on January 24, 2024, is a two-year effort led by the NSF in collaboration with multiple federal agencies, including the Department of Energy, the Department of Defense, the National Institutes of Health, and the National Institute of Standards and Technology. Private sector partners include major cloud providers such as Amazon Web Services, Microsoft Azure, Google Cloud, and Oracle Cloud Infrastructure, as well as AI hardware companies like AMD, Intel, NVIDIA (though not listed, it is a key partner), and SambaNova. The pilot provides access to high-performance computing clusters, GPUs, and AI-optimized hardware, such as AWS Trainium chips, to selected research projects.
A key feature of the pilot is the NAIRR Pilot Portal, a single entry point for researchers to request access to resources. The pilot supports three main focus areas: advancing AI research, applying AI to scientific and societal challenges, and expanding the AI workforce. For example, one funded project uses Large language models to improve weather forecasting, while another applies Deep learning to cancer imaging.
Governance and Funding
The NAIRR is governed by an interagency working group co-chaired by the NSF and OSTP. The pilot is funded through existing agency budgets, with the NSF contributing $140 million and the Department of Energy $20 million, supplemented by private sector in-kind contributions of computing time and services. A 2023 report from the OSTP estimated that a full-scale NAIRR would require $2.6 billion over six years, with annual operating costs of $500 million. The initiative has received bipartisan support in Congress, with bills introduced in both the House and Senate to authorize permanent funding.
Impact and Criticisms
Proponents argue that the NAIRR will level the playing field in AI research, allowing researchers at smaller institutions and minority-serving colleges to compete with those at MIT CSAIL, Stanford AI Lab, and BAIR (Berkeley AI Research). It also aims to foster innovation in regions outside traditional tech hubs. However, critics raise concerns about the sustainability of the pilot, the potential for bureaucratic delays, and the risk of duplicating existing private cloud services. Some researchers have noted that the pilot's allocation process is competitive and may still favor established institutions with prior experience in writing grant proposals.
Another criticism is that the NAIRR's focus on providing access to commercial cloud resources may inadvertently reinforce the dominance of large tech companies, as researchers would rely on proprietary platforms. To address this, the pilot includes partnerships with academic computing centers, such as the Texas Advanced Computing Center and the San Diego Supercomputer Center, which offer open-source software stacks.
Future Outlook
As of 2025, the NAIRR pilot is ongoing, with periodic calls for research proposals. The NSF has indicated that a full-scale NAIRR could be operational by 2026 if Congress appropriates the necessary funds. The initiative is also exploring integration with other national research infrastructures, such as the Open Science Grid and the National Research Platform. The success of the pilot will likely influence federal policy on AI infrastructure, including potential expansions to include Generative AI tools and Machine learning educational curricula.
The NAIRR represents a significant experiment in public-private partnership for scientific infrastructure. Its outcome will shape how the United States approaches equitable access to advanced computing for decades to come.
Related Initiatives
Other countries have launched similar efforts, including the European Union's AI Factories initiative and Japan's AI Computing Infrastructure. These programs share the goal of democratizing AI research but differ in governance and funding models. The NAIRR's federated approach, which combines commercial and academic resources, is unique in its scale and scope.
References
- National Science Foundation. (2024). NAIRR Pilot Launch.
- White House OSTP. (2023). Roadmap for the National AI Research Resource.
- National Security Commission on Artificial Intelligence. (2021). Final Report.